Upload trainer_0/config.json with huggingface_hub
Browse files- trainer_0/config.json +33 -0
trainer_0/config.json
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"trainer": {
|
| 3 |
+
"trainer_class": "TopKTrainer",
|
| 4 |
+
"dict_class": "AutoEncoderTopK",
|
| 5 |
+
"lr": 0.00016329931618554522,
|
| 6 |
+
"steps": 61,
|
| 7 |
+
"auxk_alpha": 0.03125,
|
| 8 |
+
"warmup_steps": 0,
|
| 9 |
+
"decay_start": null,
|
| 10 |
+
"threshold_beta": 0.999,
|
| 11 |
+
"threshold_start_step": 1000,
|
| 12 |
+
"seed": null,
|
| 13 |
+
"activation_dim": 6144,
|
| 14 |
+
"dict_size": 24576,
|
| 15 |
+
"k": 128,
|
| 16 |
+
"device": "cuda:0",
|
| 17 |
+
"layer": 4,
|
| 18 |
+
"lm_name": "blah",
|
| 19 |
+
"wandb_name": "AutoEncoderTopK",
|
| 20 |
+
"submodule_name": null
|
| 21 |
+
},
|
| 22 |
+
"buffer": {
|
| 23 |
+
"n_models": 12,
|
| 24 |
+
"d_submodule": 512,
|
| 25 |
+
"io": "out",
|
| 26 |
+
"n_ctxs": 1024,
|
| 27 |
+
"ctx_len": 128,
|
| 28 |
+
"refresh_batch_size": 512,
|
| 29 |
+
"out_batch_size": 16384,
|
| 30 |
+
"device": "cuda:0",
|
| 31 |
+
"rescale_acts": true
|
| 32 |
+
}
|
| 33 |
+
}
|